I think I’m miscommunicating, or at least overstating the degree to which I disagree, a bit – when I do my shift, I’m not eschewing grammar exercises entirely. The plan is to start up by divvying my 1.5 hours into 40 minutes grammar study, 30 reading, and 20 on WK. While I will skew more heavily towards reading once I’m satisfied that I’m familiar with the basics, I’m mostly disagreeing with the idea that reading is a reward rather than a crucial part of the learning process if only for the reason that the sheer number of words you’re exposed to increases dramatically through reading.
Bach and Mozart would like to have a word with you, at least so far as Hanon and Czerny are concerned More seriously, I think the debate is largely more about the value of non-musical keyboard exercises rather than the whether there’s any value of exercises at all. Even if you fall into the anti Hanon/Czerny/formal exercise school of thought, you do (or at least should) wind up doing targeted repetitions of difficult passages in your music, improv exercises if you will. I do think the latter is more analogous to how athletes should be practicing, ie the way that they’re going to perform come game time (though this is a bit complicated by the fact that not all athletes have the same opportunity to practice in context and so drills have to suffice), but honestly either approach has been demonstrated to work time and again so long as you’re actually getting out and playing repertoire.
As an aside, there’s an interesting paper floating around somewhere looking at baseball player performance according to how they practice. From what I recall, they set the pitching machine to one of two modes. In the first, the machine made the same pitch every time during their practice, and the baseball players got phenomenal at hitting that ball. In the second mode, the machine was made to vary the way it pitched, more similarly to how a real pitcher would vary, and players didn’t like that very much at all because they weren’t hitting home runs and whatnot. It turns out that players using this second method performed much better in actual games than when they used the first because they were used to not being thrown perfect, knock-it-outta-the-park pitches all the time. To bring all this back to Japanese, you can kind of see an analog in our use of WK. WK is a pretty tame, controlled environment, and on here my kanji accuracy is 95+%. That accuracy takes a bit of a dive out in the wild because WK is more or less throwing the same pitch every time just by nature of the system, so we need to get our uncontrolled pitches from somewhere.